Platelet counting by means of the Coulter Thrombocounter, using the Thrombofuge method for specimen preparation has been evaluated in order to determine the cause of discrepancies in results obtained in interlaboratory quality assessment surveys. The method was assessed for accuracy, precision, drift and carry-over. When the method was used, as in routine practice, some discrepancies occurred; an important source of error was identified as being due to the level of sedimented plasma layer from which samples are removed. When the method was performed strictly in accordance with the manufacturer's instructions results were generally satisfactory and comparable to those obtained with the whole blood reference method, although, even with this standardized technique, the Thrombo-fuge tended to give higher platelet counts than the reference method with counts above 200 X 10(9)/l. This discrepancy to be independent of PCV, MCV or other identifiable factors.